Dealing with statistical ruptures related to the implementation of ICD10. The case of West Germany

Researchers interested in the long-term series of mortality by cause of death always face the challenge of bridging statistical discontinuities related to the implementations of the new revisions of the WHO International Classification of Diseases (ICD). For a considerable number of countries (e.g. France, West Germany, Czech Republic, Russia and some ex-Soviet countries, UK and Netherlands), longterm continuous time series of mortality by cause of death are already available. Nevertheless, all of them end at the moment of the implementation of ICD10, which probably represents the largest qualitative and quantitative change to cause-of-death statistics in over 50 years. ICD10 came into effect in 1993, Germany adopted it for the mortality statistics in 1998. In a previous work we have reconstructed the West German cause-of-death time series from 1968 (covering the periods of ICD8 and ICD9) at the 3-digit ICD9 level. In the presented current project, we attempted to update the existing time series to the ICD10 structure and to examine the impact of introducing ICD10 on the statistical continuity. During the work, we encountered numerous problems that have led to reconsideration of the applied methodology and, eventually, to application of some non-standard procedures. We also discuss the irregular developments of selected causes of death in the period of ICD10.